An effective genetic algorithm with a critical-path-guided Giffler and Thompson crossover operator for job shop scheduling problem

Abstract

This work presents an effective genetic algorithm with a critical-path-guided Giffler and Thompson crossover operator for job shop scheduling problem with the objective of makespan minimization (GA-CPG-GT). Even though passing important traits from parents to offspring is known to be an important feature of any uniform crossover operator, most of the proposed operators adopt random exchange of genetic materials between parents; this is probably due to the fact that it is tricky to identify the genetic materials that hold the important traits. For that reason, in this work, a new selective exchange of genetic materials is proposed. In the proposed approach, at first, the genetic materials that hold the important traits are identified according to some domain specific information provided by the critical paths of the parents, and then the exchange is made on favor of them. The properties of critical path are usually utilized by the local search methods such as tabu search and simulating annealing; however, in this work, they are utilized in the global search method GA during the crossover operator. The implications of the proposed approach are investigated using the Giffler and Thompson crossover operator, which is a uniform crossover combined with the G&T algorithm. The proposed approach is tested on 55 benchmarks, with the proposed selective exchange, and without it using the random one, and also compared with other 5 similar works reported in the literature. The computational results validate the enhancements accomplished by the proposed selective exchange, and show the superiority of the proposed algorithm over the compared works in terms of solution quality, and validate its effectiveness.